Whole-Plant Growth Stage Ontology for Angiosperms and Its - TAIR

15 Pages·2006·1.33 MB·English

Bioinformatics
Whole-Plant Growth Stage Ontology for Angiosperms
and Its Application in Plant Biology1[OA]
AnuradhaPujar2,PankajJaiswal2,ElizabethA.Kellogg2,KaticaIlic2,LeszekVincent2,ShulamitAvraham2,
PeterStevens2,FelipeZapata2,LeonoreReiser3,SeungY.Rhee,MartinM.Sachs,MarySchaeffer,
LincolnStein,DoreenWare,andSusanMcCouch*
Department of Plant Breeding, Cornell University, Ithaca, New York 14853 (A.P., P.J., S.M.); Department of
Biology, University of Missouri, St. Louis, Missouri 63121 (E.A.K., P.S., F.Z.); Department of Plant Biology,
Carnegie Institution, Stanford, California 94305 (K.I., L.R., S.Y.R.); Division of Plant Sciences, University
of Missouri, Columbia, Missouri 65211 (L.V., M.S.); Cold Spring Harbor Laboratory, Cold Spring Harbor,
New York 11724 (S.A., L.S., D.W.); Missouri Botanical Garden, St. Louis, Missouri 63110 (P.S., F.Z.);
Maize Genetics Cooperation-Stock Center, Department of Crop Sciences, University of Illinois, Urbana,
Illinois 61801 (M.M.S.); and Agricultural Research Service, United States Department of Agriculture,
Washington, DC 20250 (M.M.S., M.S., D.W.)
Plant growth stages are identified as distinct morphological landmarks in a continuous developmental process. The terms
describingthesedevelopmentalstagesrecordthemorphologicalappearanceoftheplantataspecificpointinitslifecycle.The
widely differing morphology of plant species consequently gave rise to heterogeneous vocabularies describing growth and
development.Eachspeciesorfamilyspecificcommunitydevelopeddistinctterminologiesfordescribingwhole-plantgrowth
stages. This semantic heterogeneity made it impossible to use growth stage description contained within plant biology
databasestomakemeaningfulcomputationalcomparisons.ThePlantOntologyConsortium(http://www.plantontology.org)
wasfoundedtodevelopstandardontologiesdescribingplantanatomicalaswellasgrowthanddevelopmentalstagesthatcan
be used for annotation of gene expression patterns and phenotypes of all flowering plants. In this article, we describe the
developmentofagenericwhole-plantgrowthstageontologythatdescribesthespatiotemporalstagesofplantgrowthasaset
of landmark events that progress from germination to senescence. This ontology represents a synthesis and integration of
terms and concepts from a variety of species-specific vocabularies previously used for describing phenotypes and genomic
information.Itprovidesacommonplatformforannotatinggenefunctionandgeneexpressioninrelationtothedevelopmental
trajectory of a plant described at the organismal level. As proof of concept the Plant Ontology Consortium used the plant
ontologygrowthstageontologytoannotategenesandphenotypesinplantswithinitialemphasisonthoserepresentedinThe
Arabidopsis Information Resource, Gramenedatabase,andMaizeGDB.
Plant systems are complex, both structurally and thisdatawithinthecontextofanyparticularbiological
operationally, and the information regarding plant system (Gopalacharyulu et al., 2005). The ability to
development requires extensive synthesis to provide extract knowledge from historical sources and inte-
a coherent view of their growth and development. grateitwithnewinformationderivedfromglobaldata-
The difficulty of developing such a synthesis is exac- setsrequires a sophisticated approach to data mining
erbated by the deluge of new technologies such as andintegration.
high-throughputgenotyping, microarrays, proteomics, Historically, the growth and development of culti-
transcriptomics, etc., that generate large amounts of vated plants have been monitored at the whole-plant
data rapidly. The speed and magnitude of data depo- level with the help of scales of easily recognizable
sitionchallengesourabilitytorepresentandinterpret growthstages.Consequently,thereexistlargevolumes
of literature detailing growth stages for individual
1ThisworkwassupportedbytheNationalScienceFoundation plant species or closely related groups of species. For
(grantno.0321666). example,Zadok’sscale(Zadoketal.,1974)wasdevel-
2Theseauthorscontributedequallytothepaper. opedfortheTriticeaecropsandiswidelyusedtostage
3Present address: Molecular Sciences Institute, 2168 Shattuck the growth and development of cereal crops in the
Ave.,Berkeley,CA94704. UnitedStates.Theflexibilityofthisscalehasallowedit
*Corresponding author; e-mail [email protected]; fax 1–607– to be extended to other cultivated plants, and a uni-
255–6683. form code called the Biologische Bundesanstalt, Bun-
Theauthorresponsiblefordistributionofmaterialsintegraltothe
dessortenamt, and Chemical Industry (BBCH) code
findings presented in this article in accordance with the policy
was developed from it (Meier, 1997). The BBCH scale
describedintheInstructionsforAuthors(www.plantphysiol.org)is:
is quite generic and encompasses multiple crops,
SusanMcCouch([email protected]).
[OA]Open Access articles can be viewed online without a sub- including monocot and eudicot species. It offers
scription. standardized descriptions of plant development in
www.plantphysiol.org/cgi/doi/10.1104/pp.106.085720 the order of phenological appearance, and has coded
414 Plant Physiology, October 2006, Vol. 142, pp. 414–428, www.plantphysiol.org (cid:2) 2006 American Society of Plant Biologists
Plant Growth Stage Ontology
each stage for easy computer retrieval. It should be ogies (controlled vocabularies) to facilitate the anno-
noted that Arabidopsis (Arabidopsis thaliana), as a tationofgeneticinformation.Forexample,theGramene
representative of the Brassicaceae and by virtue of database(Jaiswaletal.,2006)designeditscerealgrowth
not being a cultivated species, did not have a specific stage ontology based on the stages described in the
growth stage vocabulary or scale until 2001 when standard evaluation system for rice (Oryza sativa;
Boyes et al. (2001) developed an experimental plat- INGER, 1996) and those described by Counce et al.
form describing the Arabidopsis growth stages using (2000)forrice,byZadoketal.(1974)forTriticeae(wheat
the BBCH scale. This work created a crucial semantic [Triticumaestivum],oat[Avenasativa],andbarley[Hordeum
link between Arabidopsis and cultivated plants. In vulgare]),andbyDoggett(1988)forsorghum.Exceptfor
additiontofacilitatingthedescriptionandsynthesisof the sorghum, which is a less studied crop, these species
large amounts of data within a crop species, vocabu- had fairly well-described growth staging vocabularies.
lariesliketheBBCHandZadok’sscalealsomakepos- MaizeGDB (Lawrence et al., 2005) developed a very
sible transfer of information among researchers and extensivecontrolledvocabularyfromamodifiedversion
provide a common language for comparative pur- ofthatdescribedbyRitchieetal.(1993).TheArabidopsis
poses (Counce etal., 2000). InformationResource(TAIR;Rheeetal.,2003)developed
In the post genomic era, these scales have proved the Arabidopsis growth stage ontology from the scale
inadequate to handle the deluge of information that described by Boyes et al. (2001). However, ontologies
requiredlarge-scalecomputationforcomparativeanal- createdintheseprojectsremainedrestrictedtoparticular
ysis. This called for the conversion of existing scales species or families, whereas comparative genomics re-
into ontology that have an advantage over simple quiresthatacommonstandardvocabularybeappliedto
scales because their hierarchical organization facili- abroadrangeofspecies.TheuniformBBCHscale(Meier,
tates computation across them. Terms in an ontology 1997) appeared to be a suitable model to develop a
areorganizedintheformofatree,thenodesofthetree unifiedontologysincethisscalehadalreadysynthesized
represent entities at greater or lesser levels of detail monocotandeudicotcropstagesintoasinglevocabulary.
(Smith, 2004). The branches connecting the nodes ThePlantOntologyConsortium(POC)wasinaugu-
represent the relation between two entities such that rated in 2003 for the purpose of developing common
thetermradicleemergencestageisachildofthepar- ontologies to describe the anatomy, morphology, and
ent term germination stage (Fig. 1). Individual stages growthstagesoffloweringplants(Jaiswaletal.,2005).
ofascalearethenpartsthatcanberelatedtothewhole Itsprimarytaskwastointegrateandnormalizeexist-
by their order of appearance during plant growth. ingspecies-specificontologiesorvocabulariesthathad
Each term carries a unique identifier and strictly been developed by several major databases for the
specified relationships between the terms allow sys- purpose of annotating gene expression and mutant
tematicorderingofdatawithinadatabase,thisinturn phenotype. The plant ontology (PO) is divided into
improvesinputandretrievalofinformation(Bardand two aspects. The first is the plant structure ontology
Rhee, 2004; Harris et al., 2004). (PSO), a vocabulary of anatomical terms (K. Ilic, E.A.
Consequently, several species-specific databases Kellogg, P. Jaiswal, F. Zapata, P. Stevens, L. Vincent,
converted BBCH and other scales into formal ontol- S.Avraham,L.Reiser,A.Pujar,M.M.Sachs,S.McCouch,
Figure1. Theparentandchildtermorga-
nization in the whole-plant GSO. The
solidcurvedlinesjoiningthetermsrepre-
sent IS_A relationship and the dotted
curvedlinessuggestaPART_OFrelation-
ship between the child and the parent
terms.Atermmayormaynothaveachild
term. In this example, germination IS_A
vegetativestageandfloweringIS_Arepro-
ductive stage. Similarly vegetative stage,
reproductive stage, senescence, and dor-
mancyaresubtypes(IS_A)ofwhole-plant
growthstage. Rootemergence andshoot
emergence are PART_OF the seedling
growth stage. The seedling growth stage
andimbibitionarePART_OFgermination.
Inthisimagenotallthechildrentermsare
shownforeveryparenttermintheGSO.
Plant Physiol. Vol. 142, 2006 415
Pujar et al.
M. Schaeffer, D. Ware, L. Stein, and S.Y. Rhee, unpub- tive growth and flowering terms are related by IS_A,
lisheddata),which,sinceitsreleasetothepublicdomain because flowering is a type of reproductive growth.
in 2004, has become widely used by plant genome On the other hand, seedling growth is related to its
databases(Jaiswaletal.,2005).Thesecondaspectisthe children terms radicle emergence and shoot emer-
plantgrowth and developmentalstages ontology. This gence by PART_OF, because seedling growth is com-
componentofPOisfurtherdividedintothewhole-plant prised of the two processes of radicle emergence and
growthstageontology(GSO)developedbythisproject shoot emergence(Figs. 1 and2A).
and the plant part developmental stages. This article Eachtermisgivenauniqueaccessionnumbernamed
focuses on the whole-plant GSO; we will discuss the PO:XXXXXXX where the series of X’s is a seven-digit
history,design,andapplicationsoftheGSOandshow number (Fig. 2A). Accession numbers are never re-
how it simplifies the description of a continuous and used, even when the term is retired or superseded.
complex series of events in plant development. The Obsolete terms are instead moved to a location in the
plant part developmental stages will be reviewed hierarchyunderneathatermnamedobsolete_growth_
elsewhere. and_developmental_stage(Fig.2A).Thisensuresthat
there is never any confusion about which term an ac-
cessionnumberrefersto.Eachtermalsohasahuman-
RESULTS readablenamelikeseedlinggrowth,aparagraph-length
definitionthatdescribesthecriteriaforidentifyingthe
TheGSOwasdevelopedoveraperiodoftwoyears
stage, and citations that attribute the term to a source
(2004–2006) by a team of plant biologists comprising
database,journalarticle,oranexistingstagingsystem.
systematists, molecular biologists, agronomists, plant
Many terms also have a synonym list; these are de-
breeders, and bioinformaticians. We worked to de-
scribed in moredetailbelow.
velop a set of terms to describe plant development
Our choice of the GO data model was driven by
from germination to senescence that would be valid
numerouspracticalconsiderations,foremostofwhich
across a range of morphologically distinct and evolu-
wasthefactthatthedatamodelissupportedbyarich
tionarilydistantspecies.Althoughtherateofaddition
setofdatabaseschemas,editingtools,annotationsys-
of new terms to the GSO has slowed since its initial
tems, andvisualization tools.
stagesofdevelopment,itisstillunderactivedevelop-
ment as we refine the ontology in response to user
Naming of Plant Growth Stages
input andfeedback from database curators.
Currently, the GSO has a total of 112 active terms, Thenextissuewedealtwithwashowtonameplant
eachorganizedhierarchically(Figs.1and2A)andasso- growthstages.Althoughdevelopmentinanyorganism
ciatedwithahuman-readabledefinition.Althoughwe is a continuous process, it is important to have land-
startedwithexistingsystemssuchastheBBCH(Meier, marksthatidentifydiscretemilestonesoftheprocessin
1997)aswellasthecontrolledvocabulariesdeveloped awaythatiseasilyreproducible.Extantsystemseither
forArabidopsisbyTAIR,forrice,Triticeae(wheat,oat, name growth stages according to a landmark (e.g.
and barley), and sorghum by Gramene database, and three-leaf stage) or by assigning a number or other
formaize(Zeamays)byMaizeGDB,thecurrentversion arbitrarylabeltoeachstage.Wechosetodefinegrowth
ofGSOisquitedistinctfromitspredecessors.Wewill stagesusingmorphologicallandmarksthatarevisible
first discuss the major design issues we dealt with to the naked eye (Counce et al., 2000), because such
duringthedevelopmentoftheGSO,andthendescribe descriptive terms are more intuitive, self explanatory
the structure of the GSO and its applications to real- to the users, and easy to record in an experiment. To
worldproblems.Theontologyterms,database,andgene minimize differences among species, we were able to
annotation statistics provided here are based on the describe many growth stages using measurements/
April,2006 releaseof the POC database. landmarks that are in proportion to the fully mature
state. For example, the inflorescence stages are de-
Architecture of the Ontology scribedinprogressionstartingfromthe‘‘inflorescence
justvisible,’’‘‘1/4inflorescencelengthreached,’’‘‘1/2
We chose to make use of the data model originally
inflorescence length reached,’’ to ‘‘full inflorescence
developed for the gene ontology (GO) to describe the
length reached.’’ This provides an objective measure-
GSO. This data model uses a directed acyclic graph
ment of the degree of maturation of the inflorescence
(DAG) to organize a hierarchy of terms such that the
inawaythatisnotdependentontheabsolutevalueof
most general terms are located toward the top of the
theinflorescencelength.
hierarchy while the most specific ones are located at
the bottom of the hierarchy (Figs. 1 and 2, A and B).
Synonyms
Each parent term has one or more children, and the
relationshipbetweenaparentandoneofitschildrenis Because the GSO crosses species and community
named either IS_A to indicate that the child term is a boundaries, we needed to acknowledge the fact that
specific type of the parent term, or PART_OF to each community has its own distinct vocabulary for
indicate that the child term is a component of the describing plant structures and growth stages. To ac-
parentterm(Smith,2004).Forexample,thereproduc- commodate this, we made liberal use of the GO data
416 Plant Physiol. Vol. 142, 2006
Plant Growth Stage Ontology
model’s synonym lists, which allow any GSO term to Asanexampleofasynonym,consider‘‘doughstagein
have one or more synonyms from species-specific wheat’’ and ‘‘kernel ripening in maize,’’ both of which
vocabularythatareconsideredequivalenttotheofficial essentiallyrefertothefruitripeningstage.Theseterms
termname.TheGSOcurrentlycontains997synonyms are included as synonyms to the generic (species-
takenfromseveralplantspecies.Onanaveragethere independent) GSO term ‘‘ripening’’ (PO:0007010;
areaboutninesynonymsperGSOterm(TableI).Like http://www.plantontology.org/amigo/go.cgi?view5
terms,weattributesynonymstothedatabase,literature details&show_associations5terms&search_constraint5
reference,ortextbookfromwhichtheywerederived. terms&depth50&query5PO:0007010). From the end
Figure2. TheGSOasseenontheontologybrowseravailableathttp://www.plantontology.org/amigo/go.cgi.A,Forbrowsing,
simplyclickonthe[1]iconbeforethetermnameplantgrowthanddevelopmentalstages,andthenonthe[1]nexttowhole-
plantgrowthstages(GSO).Thiswillexpandthetreebyopeningthechildrenterms.ThePOIDistheterm’saccessionnumber,
andthenumberfollowedbythetermnameisthetotalnumberofassociationsthathavebeencuratedtothegenesforagiven
term.Thisnumberwillchangedependingonthegeneproductfilterausermayhavechosen.Userscanalsogetapiechart
showingthedistributionofdataassociationstoaterm’schildrenterm.Inthisimage,thegenerallevel(toplevel)termsintheGSO
are‘‘A_Vegetativegrowth,’’‘‘B_Reproductivegrowth,’’‘‘C_Senescence,’’and‘‘D_Dormancy.’’Thesubstagesof‘‘A_Vegetative
growth’’ are ‘‘0_Germination,’’ ‘‘1_Main Shoot Growth,’’ and ‘‘2_Formation of Axillary Shoot,’’ while the substages of
‘‘B_Reproductive growth’’ are ‘‘3_Inflorescence Visible,’’ ‘‘4_Flowering,’’ ‘‘5_Fruit Formation,’’ and ‘‘6_Ripening.’’ Neither
‘‘C_Senescence’’ nor ‘‘D_Dormancy’’ currently has substages beneath them. The alphanumeric prefixes serve to make the
substagesappearintheorderinwhichtheyoccurduringtheplant’slifecycle.Ifthetemporalorderisnotdefinedconsistentlyin
allplants,thetermsmaynothavetheseprefixes.Theprefixesareusuallyabbreviationsofthetermname;forexample,LPisfor
leafproduction,SEisforstemelongation.Thenumericalportionusesdoubledigitsstartingwith01,02,andsoon.Eachofthe
substagesmayhavemorespecificstagesbeneathit.Whenatermisretiredorsuperseded,itisconsideredObsolete.Suchterms
aremovedtoalocationinthehierarchyunderneathatermnamed‘‘obsolete_growth_and_developmental_stage.’’B,Adetailed
viewofthesubstagePO:0007133,‘‘Leafproduction’’anditschildren.Childrentermsupto20leavesvisiblewereaddedto
accommodatethegrowthstagerequirementsofthemaizeplant.
Plant Physiol. Vol. 142, 2006 417
Pujar et al.
ordering is represented using either the DERIVED_
TableI. Asummaryofthenumberofsynonymsintegratedintothe
FROM, DEVELOPS_FROM, or OCCURS_AT_OR_
GSOfromeachspecies/family/source
AFTER relationship to indicate that one structure is
The integration of synonyms for Soybean and Solanaceae is in
derivedfromanotherorthatonestagefollowsanother.
progress.
However, we found these solutions to be unworkable
Species/Source No.ofSynonyms
fortheGSObecauseoftherequirementthattheontol-
Arabidopsis 93 ogy must represent growth stages across multiple
Rice 23 species. For example, consider the process of main
Maize 162
shoot growth. In the wheat plant, main shoot growth
Wheat 65
may be completed at the nine-leaf stage, while in rice
Oat 65
and maize, shoot growth may be completed at the
Barley 65
11-and20-leafstages,respectively,andthisvarieswith
Sorghum 13
BBCHandZadokscales 381 different cultivars/germplasms (Fig. 3). Transition to
Soybean 79 the subsequent stage of reproductive growth is thus
Solanaceae(mainlytomato) 51 staggered for each species, and cannot be accurately
Allthespecies 997 described by an ontology in which each stage rigidly
Averageno.ofsynonymsperGSOterm About9 followsanother.
Our compromise is to visually order the display of
user’spoint of view, thesynonyms can beused inter-
terms in a temporal and spatial fashion, but not to
changeably with the generic terms when searching
build this ordering into the structure of the ontology
databasesthat usethe PO.This means that dataasso-
itself. In practice, what we do is add alphabetic and
ciatedwiththeripeningstageofallplantsisaccessible
numeric prefixes to each term. When terms are dis-
eventoana¨ıveuser,irrespectiveofthevariabletermi-
playedtheuserinterfacetoolssortthemalphabetically
nology, diverse morphologies, and differing develop-
so that later stages follow earlier ones (Fig. 2A). This
mentaltimelinesofplantssuchaswheatandmaize.
compromise is similar to the one taken by the Dro-
By using synonyms, we were able to merge 98% of
sophiladevelopmentalstagesontology(Flybase,http://
terms from the various species-specific source ontol-
flybase.org/;OBO,2005).
ogies into unambiguous generic terms. In a few cases
As an example of how this works, we describe the
we encountered identical terms that are used by dif-
stages of leaf production using terms named ‘‘LP.01
ferentcommunitiestorefertobiologicallydistinctstages.
one leaf visible,’’ ‘‘LP.02 two leaves visible,’’ ‘‘LP.03
Weresolvedsuchcasesbyusingthesensuqualifierto
three leaves visible,’’ and so forth. When displayed
indicatethatthetermhasaspecies-specific(notgeneric)
using the ontology web browser, the terms appear in
meaning.Oneexampleofthis,describedinmoredetail
theirnaturalorder(Fig.2B).However,thereisnothing
later,is‘‘inflorescencevisible’’(tothenakedeye)versus
hard wired into the ontology that indicates that LP.01
‘‘inflorescence visible (sensu Poaceae).’’ In most plants
oneleaf visible precedes LP.02 two leaves visible.
theinflorescencebecomesvisibletothenakedeyesoon
A related issue is the observation that during plant
after it forms, whereas in Poaceae (grasses), the inflo-
maturation, multiple developmental programs can
rescenceonlybecomes visiblemuchlaterinits devel-
proceed in parallel. For instance, the processes of leaf
opment,afteremergencefromtheflagleafsheath.
production and stem elongation, although coupled,
aretemporallyoverlappingandcanproceedatdiffer-
ent relative rates among species and among cultivars
Spatiotemporal Representation
withinaspecies.Werepresentsuchprocessesasinde-
Less satisfactory is the design compromise that we pendentchildrenofamoregenericterm.Inthecaseof
reachedtorepresentthespatialandtemporalordering the previous example, both leaf production and stem
of terms. The existing plant growth scales are orga- elongation are represented as types of main shoot
nized by the temporal progression of developmental growth usingthe IS_Arelationship (Figs.1 and 2A).
events.However,theGOdatamodelpresentsunique
challenges in designing an ontology that represents
Description of the Ontology
thetemporalorderingoftermsacrossmultiplespecies
that display small butkeyvariationsinthat ordering. ThefourmaindivisionsoftheGSOare‘‘A_Vegetative
In particular, the GO data model does not have a growth,’’ ‘‘B_Reproductive growth,’’ ‘‘C_Senescence,’’
standard mechanism for representing organisms’ de- and‘‘D_Dormancy’’(Fig.2A).Asdescribedearlier,the
velopmentaltimelines.Thishasforcedeachorganism alphabeticprefixistheretoforcethesefourdivisionsto
database that has sought to represent developmental bedisplayedintheorderinwhichtheyoccurduringthe
eventsusingtheGOmodeltograpplewiththeissueof plant’slifecycleingeneral.Thesubstagesof‘‘vegetative
representingadynamicprocessinastaticrepresenta- growth’’are‘‘0_Germination,’’‘‘1_MainShootGrowth,’’
tion.Someanimalmodelorganismdatabases,suchas and‘‘2_FormationofAxillaryShoot,’’whilethesubstages
WormBase for Caenorhabditis elegans, Flybase for Dro- of‘‘reproductivegrowth’’are‘‘3_InflorescenceVisible,’’
sophila, Zfin for Zebra fish, have developed develop- ‘‘4_Flowering,’’‘‘5_FruitFormation,’’and‘‘6_Ripening.’’
mentalstageontologies(OBO,2005)inwhichtemporal Neither ‘‘C_Senescence’’ nor ‘‘D_Dormancy’’ currently
418 Plant Physiol. Vol. 142, 2006
Plant Growth Stage Ontology
Figure3. Correspondinggrowthstagesindifferentplantsandadvantagesofusingbroadandgranulartermsforannotations.In
thisexampleonecansayfloweringoccursinplantAatthesix-leafvisiblestage,inplantBatthenine-leafvisiblestage,andin
plantCatthe11-leafvisiblestage.PlantsAtoCrepresenteitherdifferentgermplasmaccessions/cultivarsofthesamespeciesor
accessions/cultivarsfromdifferentspecies.Thisnomenclatureallowstheresearchertorecordwhenageneisexpressedora
phenotypeisobservedbyfollowingthegradualprogressionoftheplant’slifecycle.Forexample,ifageneisexpressedatthesix-
leaforthefifth-internodestage,themeaningisnowclear,whileinthepast,theinformationhadtoberecordedasthefifthleaf
fromthetopoftheplant.Suchannotationrequiredthatonewaituntiltheplantcompleteditslifecycletocountthenumberof
leavesfromthetop,orthatonemakeanassumptionhowmanyleavestherewouldbeintheplant/populationusedinthestudy.
Note:thenumberofnodesandthenumberofleavesisalwayslessthanthenumberofinternodesbyone.Thearrowpointing
upwardssuggeststhatthenumbersarecountedinthatdirectioninascendingorderstartingwith1andgoingupton,wherencan
beanynumberdependingontheplant.
has substages beneath them. Again, the numeric pre- Germination(PO:0007057)
fixesarethereonlytomakethesubstagesappearina
This node in the GSO has eight children that are
logicalorder.Eachofthesubstageshasmultiple,more
broadly applicable to seed germination. The stages
specificstagesbeneathit.
under ‘‘seedling growth’’ and ‘‘shoot emergence’’ are
Although the BBCH scale (Meier, 1997) was the
not given numerical prefixes, as it is not clear which
starting point for the GSO, we have diverged from it
event precedes the other among the various species.
in many important aspects. A major difference is
Only events of seed germination were considered in
the number of top-level terms (Figs. 1 and 2A). The
this ontology, whereas the BBCH scale equates seed
BBCH scale has 10 principle stages as its top-level
germination with germination of vegetatively propa-
terms, but the GSO only has four. We collapsed four
gated annual plants and perennials such as bud
BBCH top-level stages (germination, leaf develop-
sprouting. The two processes areinfact quite distinct
ment, stem elongation, and tillering) into our top-level
in terms of organs developing at this stage, the phys-
vegetative growth term, and collapsed another six
iology, and various metabolic processes, and thus we
BBCH top-level terms (booting, inflorescence emer-
felt that combining them was inappropriate.
gence, flowering, fruit development, and ripening)
into reproductive growth. We felt justified in intro-
ducing the binning terms vegetative growth and re-
Main ShootGrowth(PO:0007112)
productive growth for several reasons; (1) to help
annotate genes that act throughout these phases; (2) Thisreferstothestageoftheplantwhentheshootis
persistentuseincurrentscientificliterature,especially undergoingrapidgrowth.Itcanbeassessedindiffer-
when the specific stage of gene action or expression entwaysdependingonthespeciesandtheinterestsof
remains unclear; and (3) they were requested by our the biologist. Plants may be equally well described in
scientificreviewerstoenhancetheimmediateutilityof termsofleavesvisibleonthemainshootorintermsof
theontology. the number of nodes detectable (Zadok et al., 1974),
We now look in more detail at some of the more andbiologistsstudyingArabidopsiscommonlyassess
importantparts of the ontology. the size of the rosette. To accommodate existing data
Plant Physiol. Vol. 142, 2006 419
Pujar et al.
associatedtothesetermswecreatedthreeinstancesof visible’’(PO:0007047),‘‘flowering’’(PO:0007026),‘‘fruit
main shoot growth, namely the ‘‘leaf production,’’ ‘‘ro- formation’’ (PO:0007042), and ‘‘ripening’’ (PO:0007010;
sette growth,’’ and ‘‘stem elongation,’’ with a strong Fig. 2A). The ‘‘inflorescence visible (sensu Poaceae)’’
recommendation to use ‘‘leaf production’’ wherever (PO:0007012) specific to grass family is an instance of
possible. the genericterm ‘‘inflorescence visible’’(PO:0007047).
Thisinturnhastwoinstances:‘‘booting’’(PO:0007014)
and ‘‘inflorescence emergence from flag leaf sheath’’
Leaf Production(PO:0007133)
(PO:0007041). As described earlier, the generic ‘‘inflo-
Leaves are produced successively so that the pro- rescence visible’’ stage is considered separate from
gressionthroughthisstagecanbemeasuredbycount- ‘‘inflorescence visible (sensu Poaceae),’’ as the former
ingthenumberofvisibleleavesontheplant(Figs.2B includes allplantswhereinflorescenceformation and
and3).Inanyspecies,leavesarealwayscountedinthe visibility coincide, while in members of the Poaceae,
same way (Meier, 1997; described in detail later). In manydevelopmentaleventsinthereproductivephase
plantsotherthanmonocotyledons,leavesarecounted start during the vegetative phase but manifest them-
when they are visibly separated from the terminal selvesasvisiblemorphologicalmarkersmuchlater.
bud. The recognition of the associated internode (be- Other stages are similar in their organization to the
low)followsthesamerule(Fig.3).Leavesarecounted existing scales, but as we continue including various
singly unless they are in pairs or whorls visibly sep- species from families Solanceae and Fabaceae, we
aratedbyaninternode,inwhichcasetheyarecounted anticipate that changes in the organization may be
as pairs or whorls. In taxa with a hypogeal type of required to accommodate them into theGSO.
germination,thefirstleafontheepicotylisconsidered
tobeleafoneandingrassesthecoleoptileisleafone.
User Interface
IntheGSOthestagesofleafproductioncontinueup
to 20 leaves/pairs/whorls of visible leaves (Fig. 2A), The GSO terms are in a simple hierarchy that is
butthiscanbeemendedtoaccommodatehighernum- intuitivetouse.TheGSOisarelativelysmallontology
bers, as new species are included. This is unlike the andhasatotalof112terms,excludingtheobsoletenode.
BBCH scale(Meier, 1997), whereonly nine leaves can It has four top nodes, 15 interior nodes (terms associ-
becountedandalltherestwouldbeannotatedtonine ated with children terms), and 88 leaf nodes (terms
leavesormore.Thiswasdonetoaccommodatetheleaf without any children terms; Fig. 2A). New terms are
development stages of maize, where depending on added based on user requests after thorough discus-
cultivars the number of leaves can be few as five or sions.AresearchercanbrowsetheGSOusingtheontol-
have20ormoreleaves.Themaizecommunityandthe ogybrowseravailableathttp://www.plantontology.org/
MaizeGDBdatabase(Lawrenceetal.,2005)useamodi- amigo/go.cgi. This is a Web-based tool for searching
fiedversionofRitchie’sscale(Ritchieetal.,1993)inwhich andbrowsingontologiesandtheirassociationstodata.
the stages of the maize plant are measured solely by It has been developed by the GO consortium (http://
counting the leaves from the seedling through the www.geneontology.org/GO.tools.shtml#in_house)and
vegetative stages,and thenodesarenot counted. modifiedtosuitourneeds.Tobrowse,clickingonthe
Stem elongation can be assessed by the number of [1]signinfrontofthetermexpandsthetreetoshow
visible nodes; this metric is commonly applied to the childrenterms (Fig.2A).Thisview providesinforma-
Triticeae,forwhichtheZadok’s(Zadoketal.,1974)or tion on the PO identification (ID) of the GSO term,
BBCH scales (Meier,1997) wereoriginallydeveloped. term name, followed by a number of associated data
Stem elongation begins when the first node becomes such as genes. For every green-colored parent term a
detectable.Thisisusuallyequivalenttonodenumber summaryofthedataassociatedtoitschildrentermsis
seven (the number varies in different cultivars), since presentedasapiechart.Theuserhasanoptiontofilter
earlier nodes are not detectable before elongation the number of associated data displayed based on
commencesinthegrasses.Boyesetal.(2001)considers species, data sources, and evidence codes. The icons
Arabidopsis rosette growth analogous to stem elon- for [i] and [p] suggest the relationship types between
gation in the grasses, and uses leaf expansion as the the parent and child term as described in the legend.
common factor linking the rosette growth and stem While browsing, a usercan click on the term name to
elongation stages. In our model, ‘‘rosette growth’’ getthedetailsatanytime(Fig.4B).Theuserswillsee
(PO:0007113)and‘‘stem elongation’’(PO:0007089)are the icon [d] fordevelops_fromrelationship type.This
treated as separate instances of sibling stages (Figs. 1 relationship type is used strictly in the PSO and not
and 2A), mainly to provide language continuity for GSO. It suggests that a plant structure develops from
users, rather than for biological reasons. another structure(Jaiswal et al., 2005).
In addition to the browse utility, users may search
byenteringthenameofatermoragene.Forexample,
ReproductiveGrowth(PO:0007130)
queryingwith‘‘germination’’resultsinthreeterms,of
Reproductive growth and its child terms are orga- which two are from the GSO section of the plant
nized a little differently from vegetative growth. Re- growthanddevelopmentstageontologyandonefrom
productive growth has four instances: ‘‘inflorescence thePSO.Toavoidgettingalargelist,usersmaychoose
420 Plant Physiol. Vol. 142, 2006
Plant Growth Stage Ontology
Figure4. AnexampleofaGSOsearchusingtheontologybrowserandsearchWebinterface.A,Ontologysearchresultsfor0
germinationbyusingtheexactmatchandtermsfilter.Tostartsearching,visitthewww.plantontology.orgWebsiteandclickon
the‘‘SearchandBrowsePlantOntology’’linkonthepagemenu.Anontologybrowserpageopensthathasasearchoptiononthe
left-handside.Typethetermnameofinterest,suchas‘‘germination’’foragenericsearchor‘‘0germination’’foranexactmatch.
Selectthetermfilterandsubmitquery.Clickonthetermnametovisitthetermdetailpageorbrowsethelineageofthistermin
theontologybyclickingthetreeiconnexttothecheckbox.B,Thetermdetailpageprovidesinformationonthetermname,
accession/ID,synonyms,definition,comments,andassociationstogenes.C,Thelistofgenesassociatedtothetermarelistedin
thebottomhalfofthetermdetailpage.Adefaultlistgivesallthegeneswitheverytypeofevidencecodeandsource.The
evidencetype,species,andsourcefilterscanbeusedtogeneratethelistasdesired.Thelistprovidesthegenesymbol,name,
source,evidence,andacitation.Thegenesymbollinkstothegenedetailpageandthesourcelinkstotheoriginalrecordinthe
contributor’sdatabase(e.g.TAIR/Gramene),theevidencecodelinkstoitsdetailsandthereferencelinkstotheoriginalcitation
referredtobythecontributorforinferringtheontologyassociationtothegene.D,Thegenedetailpageprovidesinformationon
thesymbol,name,synonym,source,alistofallthetermsintheGSOandPSO,evidence,andthecitations.Thisviewsuggests
whereandwhenageneisexpressedand/oranassociatedphenotypeisobserved.
theexactmatchoptionbeforesubmittingthequery.A externalreferencesandlinks,ifany,andtheassociated
searchfor‘‘0germination’’choosingexactmatchgives data.Theassociationsectionallowsausertoselectthe
one result (Fig. 4A). A user may browse the parents sourcedatabase,speciesname,andtheevidencecode
and children of this term by clicking on the blue- (TableII)usedtomaketheannotationtolimitthedata
coloredtreeiconandfollowingthe[1]signnexttothe displayed.Forexample,thereare138geneassociations
term name, which suggests that there are additional totheterm‘‘0germination’’(PO:0007057;Fig.4B).The
terms under this term, or simply clicking on the term listofassociateddata(Fig.4C)givesinformationabout
name ‘‘0 germination’’ for more details. The term thename,symbol,type(e.g.gene),thesource,andthe
detail page (Fig. 4B) provides information on the ID, species, in addition to the evidence used for inferring
aspectontology(plantstructureorgrowthanddevel- theassociationtotheterm.Thegenesymbolprovidesa
opment),species-specificsynonyms,ifany,definition, hyperlinktothegenedetailpage(Fig.4D),andthedata
Plant Physiol. Vol. 142, 2006 421
Pujar et al.
immediateparents(Fig.5,AandB)revealsthatmany
TableII. ListofevidencecodesforuseinannotationstoGSO
of these genes with GSO annotations in TAIR are as-
Theseareusedinbuildingtheannotationinferencesthatindicatethe
sociated to germination stages, which is a vegetative
type of experiment cited by the researcher whose data was used to
stage.Similarlythevegetativestages,particularlyfive-
determine the protein and/ortranscript expressionand phenotypeof
to six-leaf stages (children of leaf production) and
mutant(s)orquantitativetraitloci.
reproductive stages, namely the inflorescence visible
EvidenceCode Name
(sensu Poaceae), a child of inflorescence visible, fruit
IC Inferredbycurator formation, and ripening stages in therice plant areof
IDA Inferredfromdirectassay
particular importance (Fig. 5B). The Solanaceae Ge-
IEA Inferredfromelectronicannotation
nome Network (SGN) has adapted the GSO and has
IEP Inferredfromexpressionpattern
created a mapping file for Solanaceae (tomato [Lyco-
IMP Inferredfrommutantphenotype
persiconesculentum])synonymsthatisusedtoassociate
IGI Inferredfromgeneticinteraction
IPI Inferredfromphysicalinteraction theirdata. Tomato mutants areinitially being curated
ISS Inferredfromsequenceorstructuralsimilarity to these terms and, predictably, a large number of
NAS Nontraceableauthorstatement mutantswillbeassociatedtotheripeningstages(data
TAS Traceableauthorstatement not shown). As we continue to solicit data from
collaborating databases and annotate using the GSO,
weobtainaglobalviewofhowdataisassociatedwith
differentstagesof plant growth(Fig.5B).
source links to the same entry on the provider’s Web
site. This allows a user to search for extended details
thatmaynotbeprovidedinthePOCdatabase,suchas Application
information on genome location, biochemical char-
This section provides examples of genetic analyses
acterization, associations to the GOs, etc. For help at
thattypicallyusewhole-plantontogenyasafeatureof
anytime,userscanclickonthehelpmenuatthebot-
theexperimentaldesignanddataanalysis.Itindicates
tomofthebrowserpageorvisitthelinkhttp://www.
some of the difficulties of extracting spatiotemporal
plantontology.org/amigo/docs/user_guide/index.html.
informationfromthe literatureandshows theadvan-
tagesofcuratinggenomicinformationusingtheplant
Annotations to GSO
ontologies (GSO and PSO), which allow the users to
Annotation is the process of tagging snippets of query when and where a gene is assayed, expressed,
information to the genomic element by skilled biolo- or its effects become visible during the life cycle of a
gists to extract its biological significance and deepen plant. In addition the PO database supports queries
our understanding of the biological processes (Stein, such as what are the genes that are expressed during
2001).Thecuratorattributestheaddedinformationto thegerminationstageinArabidopsisandriceorshow
its sourceby the use of evidence codes (http://www. meallofthephenotypesinthereproductivestagesofa
plantontology.org/docs/otherdocs/evidence_codes.html) riceplant when mutated.
indicatingthekindofexperimentthatwascarriedout
toinfertheassociationtoaGSOterm,suchasinferred
Annotation Examples of Mutant Phenotypes
from expression pattern (IEP) involving northern,
western, and/or microarray experiment, or inferred The primary description of phenotypic data is usu-
fromdirectassay(IDA)suchasisolatedenzymeand/ allyatthewhole-plantlevelanditisrarelyastraight-
or in situ assays, etc. (Table II). The user interface has forward exercise of term-to-term association for the
queryfilteroptionstosearchforgenesannotatedwith curator. For example, characterization of dwarf mu-
agiventypeofevidencecode.Explicitspatiotemporal tants is done in different ways, most often by the leaf
information related to the whole plant is extracted or node number that is affected, counted either top
fromliteraturebyacuratoranddescribedusingterms down or bottom up; in this system the leaf and the
from the GSO. The current build of the GSO has over internode below it can be used to define the same
600genesassociatedtoitfromtheTAIRandGramene stage.Thisisdistinctfromnodevisiblestagesthatare
databases (Fig. 5A). Analysis of the data at this point lessreliable,asthefirstnodethatisvisibleisavariable
may not be entirely reflective of current research in numberin grasses(Fig. 3).
Arabidopsisandrice,asmanualcurationisadynamic An example is provided by recording of internode
and evolving process and will necessarily lag behind elongation, the main morphological feature that is
the actual state of research. In TAIR, about 130 gene affectedindwarfplants,isattributedamongothersto
associations to whole-plant growth stages carry the the effect of gibberellin and brassinosteroids (Chory,
evidence code IEP, while in the Gramene database, a 1993; Ashikari et al., 1999). Yamamuro et al. (2000)
majority of GSO annotations (about 480) carry the show that brassinosteroid plays important roles in
evidencecodeinferredfrommutantphenotype(IMP) internode elongation in rice and have characterized
and a smaller number of IEP and inferred by genetic dwarf mutants based on the specific internode that is
interaction(IGI)associations.Acloserlookatthenum- affected. In the dn-type mutant all the nodes are
ber of genes associated to various terms and their uniformly affected (the total number of nodes in a
422 Plant Physiol. Vol. 142, 2006
Plant Growth Stage Ontology
Figure5. SummaryoftheArabidopsisandricegeneannotationstotheGSO.A,Growthstage-specificgeneannotationsfrom
Arabidopsisand rice.The stagesprefixedwithA toD are the top mostcategories of thegrowth stages,namelyvegetative,
reproductive,senescence,anddormancy.Thestagesprefixedwith0to2arevegetativesubstages,andthosewith3to6are
reproductivesubstages.AllstagesmeansalltheGSOterms.B,AlistofselectedArabidopsisandricegenesannotatedtofive
specificgrowthstageterms,suggestingthecurrentstateofannotationsandnottheactualgrowthstage-specificprofile.Asimilarlist
canbegeneratedtogetgrowthstage-specificgeneexpressionprofilesforagivenspecies.Incolumns2and3,thenumbers(written
inbold)appearingbeforetheparenthesesarethetotalnumberofgeneannotations;species-specificgenesarewritteninitalics.
given mature rice plant). However, in the nl-type to the researcher and curator as well as the user (Fig.
mutant, only the fourth internode is affected, while 3). Currently by using the IMP filter, more than 500
in the case of the sh-type mutant, only the first inter- genes annotated to different growth stages are avail-
node is affected. However, in this case,the authors of ablein the PO database.
the study number the internodes from top down (the
uppermost internode below the panicle is the first
Cross-Database Comparison of Gene Annotations
internode).TobeconsistentwiththeGSO,thesenum-
bershavetobeconvertedtotheappropriateleaf/node Almost all organismal databases are mutually ex-
countingfromthebaseoftheshoot(Fig.3).Thishasto clusive and provide little or no overlap in their sche-
be achieved by the curator’s personal knowledge of maswithotherdatabases.Thustheycatertoexclusive
the plant, from legacy information available for the user communities. To illustrate how the use of ontol-
speciesandgermplasmaccession,orbycontactingthe ogies can overcome database interoperability prob-
authors. Unlike the above example, generally leaves lems, we compare the related processes of flowering
arecountedfrombelowandthecuratorextractsinfor- time in Arabidopsis and heading date in rice (Fig. 6).
mation from statements such as when the plant is at Thegenenetworkunderlyingthephotoperiodicflow-
the three-leaf stage. This permits an immediate visu- ering response involves photoreceptors, circadian
alizationofthemorphologicalappearanceoftheplant clock systems, and floral regulator genes (Yanovsky
Plant Physiol. Vol. 142, 2006 423